and How to Give a Good Talk - College of...
Transcript of and How to Give a Good Talk - College of...
How to Write a Good Paperand
How to Give a Good Talk
Liang HuangQueens College and Graduate Center
The City University of New York
(some of the material also applies to “how to teach a class”)
Why I am giving this talk...
• not because I am a good writer (in fact I’m not)...
• but because I was a terrible writer!
2
David Chiang (usc)paper writing
Kevin Knight (usc)presentation
Ed Hovy (usc)proposal writing
Shanghua Teng (usc)teaching
How I learned writing
• ’03: knew nothing about writing (though I wrote several papers in China)
• ’04-5: all I wrote was crap; David turned them into beauty
• ’06-7: some progress by writing, writing, and writing...
• one of the reviews for a submission with David (rejected)
• “in general this paper is written with admirable clarity, except for it doesn’t seem to be written by a single author or with the same level of discretion...” (this made me not sad about the rejection... :P)
• turns out David had revised all but one section
• first single-author paper (not a good one, but great practice)
• ’08 and on: all my submissions got 4 or 5 in “clarity”3
How I learned writing
4
• fallacy: students learn to write mainly from advisors
• truth: learn from anybody whom you can learn from
• I learned writing mainly from...
• and from writing seminars of...
• and from the slides by...K. KnightD. Chiang
D. GildeaB. PierceL. Saul S. Peyton-Jones
the rest of the talkis largely based on Simon PJ’s slides.
This is NOT an English class
• writing is not about language, but about logic
• writing is equally hard for both native and non-native speakers of English
• a bad paper is bad in any language
• different levels of writing
• high-level (paper): global shape, logic, argument, style
• mid-level (discourse): coherence within a paragraph
• low-level (sentences): ordering of words and phrases
• lowest-level (words): word choice, grammar5
First Principle: Audience-Centric
• always have your audience (the reader) in mind!
• writing is communication, NOT self-expression!
• reader-centric attitude, not self-centric
6
Writing papers: model 1
Idea Do research Write paper
Writing papers: model 2
Idea Do research Write paper
Idea Write paper Do research
n Forces us to be clear, focusedn Crystallises what we don’t understandn Opens the way to dialogue with others:
reality check, critique, and collaboration
J. Eisner
Do not be intimidated
Write a paper, and give a talk, about
any idea, no matter how weedy and insignificant it
may seem to you
Fallacy You need to have a fantastic idea before you can write a paper. (Everyone else seems to.)
Do not be intimidated
Write a paper, and give a talk, about any idea, no matter how insignificant it may
seem to you
n Writing the paper is how you develop the idea in the first place
n It usually turns out to be more interesting and challenging that it seemed at first
LH: talk, write as early as you can; don’t wait until you feel ready;
it doesn’t mean you have to publish it.
The purpose of your paper
Papers communicate ideas
n Your goal: to infect the mind of your reader with your idea, like a virus
n Papers are far more durable than programs (think Mozart)
The greatest ideas are (literally) worthless if you keep them to
yourself
The purpose of your paper is not...
To describe the WizWoz system
§ Your reader does not have a WizWoz
§ She is primarily interested in re-usable brain-stuff, not executable artefacts
Your narrative flow
n Here is a problemn It’s an interesting problemn It’s an unsolved problemn Here is my idean My idea works (details, data)n Here’s how my idea compares to other
people’s approaches
I wish I knew how to solve
that!
I see how that works. Ingenious!
Structure (conference paper)
n Title (1000 readers)n Abstract (4 sentences, 100 readers)n Introduction (1 page, 100 readers)n The problem (1 page, 10 readers)n My idea (2 pages, 10 readers)n The details (5 pages, 3 readers)n Related work (1-2 pages, 10 readers)n Conclusions and further work (0.5 pages)
The abstract
n I usually write the abstract lastn Used by program committee members to
decide which papers to readn Four sentences [Kent Beck]
1. State the problem2. Say why it’s an interesting problem3. Say what your solution achieves4. Say what follows from your solution
Example
1. Many papers are badly written and hard to understand
2. This is a pity, because their good ideas may go unappreciated
3. Following simple guidelines can dramatically improve the quality of your papers
4. Your work will be used more, and the feedback you get from others will in turn improve your research
Structure
n Abstract (4 sentences)n Introduction (1 page)n The problem (1 page)n My idea (2 pages)n The details (5 pages)n Related work (1-2 pages)n Conclusions and further work (0.5 pages)
The introduction (1 page)
1. Describe the problem2. State your contributions...and that is all
ONE PAGE!
LH: this is the hardest part of writing!need to convey: importance and hardness
abstract:
intro:
LH Method for Stating the Problem
• intro = “your slightly biased view of the history” [N. Dinesh]
• need to convey: importance and depth
• this is an important problem
• the dominant solution is good in A
• but bad in B (and B is important)
• the alternative solution is good in B
• but bad in A
• Q: how to combine their merits?? a hard problem!20
A Bs1s2
new
+ -
- +
+ +
21
exam
ple
from
Hua
ng (2
008)
State your contributions
n Write the list of contributions firstn The list of contributions drives the
entire paper: the paper substantiates the claims you have made
n Reader thinks “gosh, if they can really deliver this, that’s be exciting; I’d better read on”
State your contributions
Bulleted list of
contributions
Do not leave the reader to guess what your contributions are!
Contributions should be refutableNO! YES!
We describe the WizWoz system. It is really cool.
We give the syntax and semantics of a language that supports concurrent processes (Section 3). Its innovative features are...
We study its properties We prove that the type system is sound, and that type checking is decidable (Section 4)
We have used WizWoz in practice
We have built a GUI toolkit in WizWoz, and used it to implement a text editor (Section 5). The result is half the length of the Java version.
No “rest of this paper is...”
n Not:
n Instead, use forward references from the narrative in the introduction. The introduction (including the contributions) should survey the whole paper, and therefore forward reference every important part.
“The rest of this paper is structured as follows. Section 2 introduces the problem. Section 3 ... Finally, Section 8 concludes”.
Structure
n Abstract (4 sentences)n Introduction (1 page)
nRelated workn The problem (1 page)n My idea (2 pages)n The details (5 pages)n Related work (1-2 pages)
No related work yet!
Related work
Your reader Your ideaWe adopt the notion of transaction from Brown [1], as modified for distributed systems by White [2], using the four-phase interpolation algorithm of Green [3]. Our work differs from White in our advanced revocation protocol, which deals with the case of priority inversion as described by Yellow [4].
No related work yet
n Problem 1: the reader knows nothing about the problem yet; so your (carefully trimmed) description of various technical tradeoffs is absolutely incomprehensible
n Problem 2: describing alternative approaches gets between the reader and your idea
I feel tired
I feel stupid
LH: Two Types of Previous Work
• essential background
• the previous work that your work builds upon
• or improve upon (“shoulders of giants”)
• => intro (w/o which readers can’t understand your work)
• related work: other previous work that is just related to yours
• having them doesn’t change the understanding your work
• simple criteria: can readers understand my work w/o A?
29
your work (2) related work (3)
essential background (1)essential background (1)
Structure
n Abstract (4 sentences)n Introduction (1 page)n The problem (1 page)n My idea (2 pages)n The details (5 pages)n Related work (1-2 pages)n Conclusions and further work (0.5 pages)
Presenting the idea
n Explain it as if you were speaking to someone using a whiteboard
n Conveying the intuition is primary, not secondary
n Once your reader has the intuition, she can follow the details (but not vice versa)
n Even if she skips the details, she still takes away something valuable
The payload of your paper
Introduce the problem, and your idea, using
EXAMPLESand only then present the
general case
Using examples
Example right away
The Simon PJ question: is there
any typewriter font?
Examples and Illustrations
34
Examples and Illustrations
35
The details: evidence
n Your introduction makes claimsn The body of the paper provides evidence to support each claim
n Check each claim in the introduction, identify the evidence, and forward-reference it from the claim
n Evidence can be: analysis and comparison, theorems, measurements, case studies
Structure
n Abstract (4 sentences)n Introduction (1 page)n The problem (1 page)n My idea (2 pages)n The details (5 pages)n Related work (1-2 pages)n Conclusions and further work (0.5 pages)
Related work
Fallacy To make my work look good, I have to make other people’s work look bad
The truth: credit is not like money
Giving credit to others does not diminish the credit you get from your paper
§ Warmly acknowledge people who have helped you
§ Be generous to the competition. “In his inspiring paper [Foo98] Foogle shows.... We develop his foundation in the following ways...”
§ Acknowledge weaknesses in your approach
Credit is not like money
Failing to give credit to others can kill your paper
If you imply that an idea is yours, and the referee knows it is not, then either
§ You don’t know that it’s an old idea (bad)
§ You do know, but are pretending it’s yours (very bad)
Structure
n Abstract (4 sentences)n Introduction (1 page)n The problem (1 page)n My idea (2 pages)n The details (5 pages)n Related work (1-2 pages)n Conclusions and further work (0.5 pages)
Conclusions and further work
n Be brief.
The process of writing
The process
n Start early. Very early. n Hastily-written papers get rejected.n Papers are like wine: they need time to
mature
n Collaborate
n Use CVS to support collaboration
Getting help
n Experts are goodn Non-experts are also very goodn Each reader can only read your paper for the
first time once! So use them carefullyn Explain carefully what you want (“I got lost
here” is much more important than “Jarva is mis-spelt”.)
Get your paper read by as many friendly guinea pigs as possible
Listening to your reviewers
Treat every review like gold dustBe (truly) grateful for criticism as
well as praise
This is really, really, really hard
But it’s really, really, really, really, really, really,
really, really, really, really important
Listening to your reviewers
n Read every criticism as a positive suggestion for something you could explain more clearly
n DO NOT respond “you stupid person, I meant X”. Fix the paper so that X is apparent even to the stupidest reader.
n Thank them warmly. They have given up their time for you.
Language and style
Visual structure
Visual Structure
50
Use the active voice
NO YESIt can be seen that... We can see that...
34 tests were run We ran 34 tests
These properties were thought desirable
We wanted to retain these properties
It might be thought that this would be a type error
You might think this would be a type error
The passive voice is “respectable” but it DEADENS your paper. Avoid it at all costs.
“We” = you and the reader
“We” = the authors
“You” = the reader
Newton uses the active voice!
n I held the Prism. n I looked through the Prismn I stopt the Prismn I observed the length of its refracted
Imagen I removed the Prism out of the Sun’s
Light and looked
52
Sir Isaac Newton (1704). Optics.
Use simple, direct language
NO YESThe object under study was
displaced horizontally The ball moved sideways
On an annual basis Yearly
Endeavour to ascertain Find out
It could be considered that the speed of storage reclamation left something to be desired
The garbage collector was really slow
Resources for the Writing Part• writing resources: http://www.cis.upenn.edu/~lhuang3/writing/
• high-level (language-independent)
• Simon Peyton-Jones: How to Write a Research Paper
• Mark-Jan Nederhof: Common Pitfalls in Academic Writing
• low-level (language-specific -- use NLP!)
• Gopen & Swan: The Science of Scientific Writing
• Williams: STYLE: Clarity and Grace series
• Strunk and White: The Elements of Style
• Cook: Line by Line
54
How to give a good research talk
Simon Peyton JonesMicrosoft Research, Cambridge
1993 paper joint with John Hughes (Chalmers),
John Launchbury (Oregon Graduate Institute)
How to give a good research talk
Simon Peyton JonesMicrosoft Research, Cambridge
1993 paper joint with John Hughes (Chalmers),
John Launchbury (Oregon Graduate Institute)
Do it! Do it! Do it!
Good papers and talks are a fundamental part of research excellence
§ Invest time
§ Learn skills
§ Practice
Write a paper, and give a talk, about any idea,
no matter how weedy and insignificant it may seem to you
Giving a good talk
This presentation is about how to give a good research talk
§ What your talk is for
§ What to put in it (and what not to)
§ How to present it
What your talk is for
Your paper = The beef
Your talk = The beef advertisment
Do not confuse the two
The purpose of your talk…
..is not:
§ To impress your audience with your brainpower
§ To tell them all you know about your topic
§ To present all the technical details
The purpose of your talk…
..but is:
§ To give your audience an intuitive feel for your idea
§ To make them foam at the mouth with eagerness to read your paper
§ To engage, excite, provoke them
Your ideal audience…
The audience you would like
§ Have read all your earlier papers
§ Thoroughly understand all the relevant theory of cartesian closed endomorphic bifunctors
§ Are all agog to hear about the latest developments in your work
§ Are fresh, alert, and ready for action
Your actual audience…The audience you get
§ Have never heard of you
§ Have heard of bifunctors, but wish they hadn’t
§ Have just had lunch and are ready for a doze
Your mission is to
WAKE THEM UPAnd make them glad they did
What to put in
What to put in
1. Motivation (20%)2. Your key idea (80%)3. There is no 3
MotivationYou have 2 minutes to engage your audience
before they start to doze
§ Why should I tune into this talk?
§ What is the problem?
§ Why is it an interesting problem?Example: Java class files are large (brief figures), and get sent over the network. Can we use language-aware compression to shrink them?
Example: synchronisation errors in concurrent programs are a nightmare to find. I’m going to show you a type system that finds many such errors at compile time.
Your key idea
If the audience remembers only one thing from your talk, what should it be?
§ You must identify a key idea. “What I did this summer” is No Good.
§ Be specific. Don’t leave your audience to figure it out for themselves.
§ Be absolutely specific. Say “If you remember nothing else, remember this.”
§ Organise your talk around this specific goal. Ruthlessly prune material that is irrelevant to this goal.
Narrow, deep beats wide, shallow
No
Yes
§Avoid shallow overviews at all costs§Cut to the chase: the technical “meat”
Your main weapon 1
Examples are your main weapon
§ To motivate the work§ To convey the basic intuition§ To illustrate The Idea in action§ To show extreme cases§ To highlight shortcomings
When time is short, omit the general case, not the example
Liang Huang (USC)
LH: Your main weapon #2
70
Visualization!a picture is worth a
thousand words!
beam searchgreedy search
Liang Huang (USC)
Example: Dynamic Programming• each state => three new states (shift, l-reduce, r-reduce)
• key idea of DP: share common subproblems
• merge equivalent states => polynomial space
71(Huang and Sagae, 2010)
Liang Huang (USC)
Example: Dynamic Programming• each state => three new states (shift, l-reduce, r-reduce)
• key idea of DP: share common subproblems
• merge equivalent states => polynomial space
72(Huang and Sagae, 2010)
Liang Huang (USC)
Real Life Analogy: Lebesgue Integral
73
• Riemann Integral (Newton-Leibniz Style)
• intuitive, but left many important functions unintegrable
• Lebesgue Integral
• greatly extended the domain of integrable functions
• Real Life Analogy
Lebesgue to Paul Montel:I have to pay a certain sum, which I have collected in my pocket. I take the bills and coins out of my pocket and give them to the creditor in the order I find them until I have reached the total sum. This is the Riemann integral. But I can proceed differently. After I have taken all the money out of my pocket I order the bills and coins according to identical values and then I pay the several heaps one after the other to the creditor. This is my integral.—Source: (Siegmund-Schultze 2008)
Liang Huang (USC)
Real Life Analogy: Public-Key
74
private key
public key
What to leave out
Outline of my talk§ Background
§ The FLUGOL system
§ Shortcomings of FLUGOL
§ Overview of synthetic epimorphisms
§ π-reducible decidability of the pseudo-curried fragment under the Snezkovwski invariant in FLUGOL
§ Benchmark results
§ Related work
§ Conclusions and further work
No outline!
“Outline of my talk”: conveys near zero information at the start of your talk
§ But you can put up an outline for orientation after your motivation
§ …and signposts at pause points
Related work
[PMW83] The seminal paper
[SPZ88] First use of epimorphisms
[PN93] Application of epimorphisms to wibblification
[BXX98] Lacks full abstraction
[XXB99] Only runs on Sparc, no integration with GUI
Do not present related work
But
§ You absolutely must know the related work; respond readily to questions
§ Acknowledge co-authors (title slide), and pre-cursors (as you go along)
§ Do not disparage the opposition
§ X’s very interesting work does Y; I have extended it to do Z
Technical detail
Omit technical details
§ Even though every line is drenched in your blood and sweat, dense clouds of notation will send your audience to sleep
§ Present specific aspects only;refer to the paper for thedetails
§ By all means have backup slides to use in response to questions
Presenting your talk
Polish your slides the night before
(…or at least, polish it then)
Your talk absolutely must be fresh in your mind
§ Ideas will occur to you during the conference, as you obsess on your talk during other people’s presentations
§ Do not use typeset slides, unless you have a laptop too
§ Handwritten slides are fine
§ Use permanent ink
§ Get an eraser: toothpaste does not work
How to present your talk
By far the most important thing is to
be enthusiastic
Enthusiasm
§ If you do not seem excited by your idea, why should the audience be?
§ It wakes ‘em up
§ Enthusiasm makes people dramatically more receptive
§ It gets you loosened up, breathing, moving around
Liang Huang (USC)
LH: Be Fun -- Three Jokes Rule
• Include as many relevant jokes as possible
• three jokes rule
• one at the beginning (motivation)
• one at the middle (to wake up people)
• and one at the end
• especially important in job talks!
86
Liang Huang (USC)
Relevant Jokes: Translation Errors
87
liang’s rule: if you see “X carefully” in China,
just don’t do it.
Liang Huang (USC)
Relevant Jokes: Translation Errors
88
Liang Huang (USC)
Relevant Jokes: Translation Errors
89
clear evidence that MT is used in real life.
Liang Huang (USC)
LH: Use a wireless presenter
• wireless click + laser pointer + [USB disk]
• smoothes your transition!
90
Liang Huang (USC)
Face the audience• avoid looking and pointing at your laptop
• look at the audience (70%), screen (25%), and laptop (5%)
• where is the place to stand as a speaker?
91
Finishing
Absolutely without fail, finish on time
§ Audiences get restive and essentially stop listening when your time is up. Continuing is very counter productive
§ Simply truncate and conclude
§ Do not say “would you like me to go on?” (it’s hard to say “no thanks”)
There is hope
The general standard is so low that you don’t
have to be outstanding to stand out
You will attend 50x as many talks as you give. Watch other people’s talks intelligently, and pick up
ideas for what to do and what to avoid.
Liang Huang (USC)
Conclusion: Technical Communication
94
interaction
technical details
almost zero a little or a lot a whole lot
needed not needed needed
hardest easiestdifficulty